多传感器信息融合的前方车辆检测

为提升辅助驾驶系统对于道路环境中车辆的感知能力，通过机器视觉与毫米波雷达信息融合技术对前方车辆进行了检测。融合系统中对摄像头和毫米波雷达进行了联合标定，借助三坐标测量仪确定两者的数据转换的关系，优化了深度学习算法SSD的候选框，提高了车辆的检测速度，选用长焦和短焦两种摄像头进行前方图像采集，并将两者重合图像进行融合，提升了前方小目标图像的清晰度，同时对毫米波雷达数据进行了处理，借助雷达模拟器确定合适阈值参数实现对车辆目标的有效提取，根据雷达有效目标数据对摄像头采集的图像进行选择与建立感兴趣区域，通过改进的SSD车辆识别算法对区域中的车辆进行检测，经过测试，车辆的检测准确率最高达到95.3%，单帧图像平均处理总时间为32 ms，该算法提升系统前方车辆检测的实时性和环境适应性。

Front vehicle detection based on multi-sensor information fusion

In order to improve the ability of Advanced Driving Assisted System (ADAS) to perceive vehicles in the road environment, the information fusion algorithm of machine vision and millimeter wave radar was proposed to detect front vehicles in this paper. Firstly, the camera and millimeter wave radar were jointly calibrated to determine their conversion formula using the coordinate measuring machine in the fusion system. The candidate frame of SSD for deep learning algorithm was optimized to improve the speed of vehicle detection, while long focus camera and short focus camera were selected for two front images acquisition, the overlapped images were fused to improve sharpness of small target image ahead. The appropriate threshold parameters of radar data were determined by radar simulator and the effective vehicle target was extracted. According to these effective target data, the image collected by the camera was selected and the region of interest was established. Vehicles in the selection region were detected with the improved SSD algorithm. In the test, the vehicle detection rate is 95.3%, and the total processing time for single frame image is 32 ms. It proves that the algorithm can help ADAS system to archieve vehicle detetcion with higher real-time and environmental adaptability.